The Web server (running the Web site) is currently unable to handle the HTTP request due to a temporary overloading or maintenance of the server. The implication is that this is a temporary condition which will be alleviated after some delay. Some servers in this state may also simply refuse the socket connection, in which case a different error may be generated because the socket creation timed out.
Fixing 503 errors
The Web server is effectively ?closed for repair?. It is still functioning minimally because it can at least respond with a 503 status code, but full service is impossible i.e. the Web site is simply unavailable. There are a myriad possible reasons for this, but generally it is because of some human intervention by the operators of the Web server machine. You can usually expect that someone is working on the problem, and normal service will resume as soon as possible.
Please contact the system operators of the Web site (e.g. your ISP) to determine why the service is down. They will be in a much better position to help you than we are for this type of error.
The Amazon ELB will return this error if there are no healthy instances. Sites may be misconfigured, so the ELB healthcheck will fail, which causes the ELB to take the servers out of rotation. With zero healthy sites, the ELB returns 503 Service Unavailable: Back-end server is at capacity.